Operating room infection management system based on big data

Abstract

The application discloses a surgery room infection management system based on big data, which comprises a data acquisition module, a preoperative surgery room infection risk assessment module, an intraoperative surgery room infection dynamic monitoring module, a postoperative surgery room infection terminal assessment module and a surgery room infection intelligent management module. The application relates to the technical field of data processing, in particular to a surgery room infection management system based on big data. The application innovatively combines preoperative, intraoperative and postoperative surgery room infections to realize dynamic monitoring and whole-process control. The application introduces a segmented convergence factor and an iterative level reverse learning optimization population strategy to improve the optimization algorithm, thereby improving the model hyperparameter optimization precision and the accuracy of model output results. The application designs a bidirectional convolution path, combines a gated recurrent unit and simultaneously adopts a model semi-supervised loss function with a supervised loss, a consistency regularization loss and a pseudo label loss, thereby significantly improving the precision and real-time performance of the monitoring model.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, specifically to an operating room infection management system based on big data. Background Technology

[0002] Operating room infection management system is an intelligent infection prevention and control management system designed for operating room scenarios. By collecting multi-source heterogeneous data from the operating room and utilizing big data processing and data analysis technologies, it assesses and dynamically monitors operating room infection-related risks, thereby generating operating room infection risk management results. It can provide real-time early warning and decision support for operating room managers, realize intelligent management of operating room infection risks, help improve operating room operational safety and management efficiency, and reduce the risk of cross-contamination.

[0003] However, traditional operating room infection management is mostly based on single-stage control, lacking a coherent system encompassing preoperative risk prediction, intraoperative dynamic control, and postoperative final review. This results in technical issues such as insufficient accuracy in operating room infection risk assessment and monitoring. Existing hyperparameter search algorithms suitable for evaluation models suffer from slow convergence speeds and a tendency to get trapped in local optima, leading to insufficient hyperparameter optimization accuracy, poor model output accuracy, and low evaluation efficiency. Existing models for intraoperative operating room infection monitoring rely solely on the extraction of single temporal features, lack effective bidirectional temporal dependency capture, and are overly dependent on labeled data. However, the large amount of unlabeled data in the operating room is difficult to utilize, resulting in poor accuracy, real-time performance, and generalization ability of the infection monitoring model's output. Summary of the Invention

[0004] To address the above issues and overcome the shortcomings of existing technologies, this invention provides a big data-based operating room infection management system. Traditional operating room infection management often involves single-stage control, lacking a coherent system encompassing preoperative risk prediction, intraoperative dynamic control, and postoperative final review. This results in insufficient accuracy in operating room infection risk assessment and monitoring. This solution innovatively combines a preoperative operating room infection risk assessment module, an intraoperative operating room infection dynamic monitoring module, and a postoperative operating room infection final assessment module to form a complete, closed-loop infection risk management system. This ensures that infection risks at each stage can be monitored in real-time, effectively controlled, and promptly rectified, thereby achieving comprehensive infection risk management. Dynamic monitoring and full-process control improve the early warning capability of infection risks and enhance the dynamism and real-time nature of operating room infection management. It enables timely detection and handling of potential risks at each stage of surgery, improving the comprehensiveness of operating room infection control and achieving intelligent and automated management of infection prevention and control. Addressing the technical problems of slow convergence speed and susceptibility to local optima in existing hyperparameter search algorithms suitable for evaluation models, leading to insufficient hyperparameter optimization accuracy, poor model output accuracy, and low evaluation efficiency, this solution innovatively introduces a piecewise convergence factor and an iterative back-learning optimization population strategy to improve the optimization algorithm. The piecewise nonlinear convergence factor can dynamically balance the algorithm's global exploration and... The local development capability accelerates the convergence process of the algorithm towards the optimal hyperparameter combination. The iterative back-learning population optimization strategy can continuously improve population diversity throughout the entire iteration cycle of the algorithm, significantly enhancing the global optimization ability and iterative stability, improving the efficiency and reliability of hyperparameter search, thereby improving the model's hyperparameter optimization accuracy and the accuracy of the model output results, and ultimately achieving efficient decision-making for intelligent management of operating room infections. This addresses the shortcomings of existing models applicable to intraoperative operating room infection monitoring, which rely solely on the extraction of single temporal features, lack effective bidirectional temporal dependency capture, and overly depend on labeled data. However, the large amount of unlabeled data in the operating room is difficult to utilize, leading to inaccuracies in the output results of infection monitoring models. To address the technical issues of poor accuracy, real-time performance, and generalization ability, this solution innovatively proposes a bidirectional convolutional path combined with a gated recurrent unit. It also employs a semi-supervised loss function that integrates supervised loss, consistency regularization loss, and pseudo-label loss for semi-supervised model training. This improves the ability to extract temporal features from operating room infection risk during surgery, overcomes the dependence of traditional models on fully labeled data, and solves the problem of underutilization of unlabeled data in operating room infection monitoring. It significantly improves the accuracy and real-time performance of the monitoring model, enhances the comprehensiveness and stability of infection risk monitoring, and improves the model's robustness and generalization ability, ultimately achieving intelligent and automated operating room infection monitoring.

[0005] The technical solution adopted by the present invention is as follows: The operating room infection management system based on big data provided by the present invention includes a data acquisition module, a preoperative operating room infection risk assessment module, an intraoperative operating room infection dynamic monitoring module, a postoperative operating room infection terminal assessment module, and an operating room infection intelligent management module; The data acquisition module is used to obtain optimized data for operating room infection management; The preoperative operating room infection risk assessment module is used to establish a preoperative operating room infection risk assessment model. It adopts an improved optimization algorithm that introduces a piecewise convergence factor and an iterative back learning optimization population strategy to obtain the optimal hyperparameter combination of the model and adjust the model hyperparameters. Based on the model after hyperparameter adjustment, the real-time preoperative operating room infection risk assessment result is obtained. The intraoperative operating room infection dynamic monitoring module is used to extract key bidirectional temporal features of intraoperative infection through bidirectional convolutional paths, and to construct an intraoperative operating room infection monitoring model by combining gated recurrent units and attention mechanisms. The model is semi-supervised by using a model semi-supervised loss function that integrates supervised loss, consistency regularization loss, and pseudo-label loss. Based on the trained model, the real-time intraoperative operating room infection risk monitoring results are obtained. The postoperative operating room infection terminal assessment module is used to obtain real-time postoperative operating room infection terminal assessment results; The intelligent management module for operating room infection is used to intelligently manage the infection status of the operating room based on real-time results before, during, and after surgery.

[0006] Furthermore, the data acquisition module specifically obtains raw operating room infection management data by performing data acquisition operations, and performs data preprocessing on the raw operating room infection management data to obtain optimized operating room infection management data; The raw data for operating room infection management includes historical infection management data and real-time infection management data; Both the historical infection management data and the real-time infection management data include preoperative operating room infection management data, intraoperative operating room infection management data, and postoperative operating room infection management data; The historical infection management data also includes historical operating room infection management results; The data preprocessing specifically involves cleaning, normalizing, and selecting features from the raw data to obtain optimized operating room infection management data.

[0007] Furthermore, the preoperative operating room infection risk assessment module specifically includes the following steps: A preoperative operating room infection risk assessment model was established. Specifically, the model was built based on a multilayer perceptron neural network. Preoperative operating room infection management data from historical infection management optimization data was used as training data for the model to obtain the trained preoperative operating room infection risk assessment model. Constructing a model hyperparameter optimization algorithm, specifically, obtaining the optimal combination of hyperparameters for the model through an improved optimization algorithm; including the following steps: Initialize the population of search individuals by encoding the model's hyperparameters into search individual position vectors and generating M search individual position vectors through random initialization to form the initial search individual population; The fitness calculation of individual search involves calculating the fitness value of individual search in the population, using the model performance built based on the hyperparameter combination corresponding to the position of the individual search as the fitness value of the individual search. To calculate the piecewise convergence factor, firstly, the split point between the nonlinear and linear segments is calculated based on the number of iterations, and then the piecewise convergence factor is calculated using the split point as the boundary. The search individual location update specifically involves updating the search individual location for each search individual based on a random number and search control parameters, according to different scenarios. Iterative back-learning optimizes the population. Specifically, for the updated search individuals, back-search individuals are generated based on the search space boundary, resulting in a back-search population. The original population and the back-search population are then merged to obtain a mixed population of size 2M. Finally, the fitness values ​​of all search individuals in the mixed population are calculated, and after sorting them by fitness, the M search individuals with the best fitness are selected as... The initial population is iterated, and the individual with the best fitness value is selected for the search, and the global optimal position of the search individual is updated. The search iteration terminates when the fitness value corresponding to the global optimal position of the search individual is higher than the fitness threshold or when the number of iterations reaches the maximum number of iterations. The global optimal position of the search individual specifically refers to the optimal combination of hyperparameters of the model. Model hyperparameter tuning specifically involves obtaining the optimal hyperparameter combination of the preoperative operating room infection risk assessment model based on the model hyperparameter optimization algorithm, adjusting the hyperparameters of the trained preoperative operating room infection risk assessment model, and obtaining the optimal preoperative operating room infection risk assessment model. Real-time preoperative operating room infection risk assessment involves inputting preoperative operating room infection management data from real-time infection management optimization data into the optimal preoperative operating room infection risk assessment model to obtain real-time preoperative operating room infection risk assessment results.

[0008] Furthermore, the intraoperative operating room infection dynamic monitoring module specifically includes the following steps: The construction of an intraoperative operating room infection monitoring model includes the following steps: The extraction of key bidirectional temporal features of intraoperative infection involves designing a bidirectional convolutional neural network to extract temporal features of both positive and negative infection risk, and then fusing these features to obtain the key bidirectional temporal features of intraoperative infection. This includes the following steps: The design of the forward convolution path involves taking the dynamic monitoring feature set of intraoperative operating room infection as input, performing sliding convolution calculation along the forward temporal direction of the operation time, and obtaining the forward infection risk temporal features through activation function and max pooling operation. The design of the reverse convolution path is as follows: the feature set of dynamic monitoring of intraoperative operating room infection is subjected to sliding convolution calculation along the reverse temporal direction of the flip, and then the reverse pooling feature is obtained through activation function and max pooling operation. The reverse pooling feature is then flipped along the time axis again to obtain the reverse infection risk temporal feature. Bidirectional feature fusion specifically involves aligning the temporal features of positive infection risk and reverse infection risk in dimensions, and then splicing and fusing them in the feature channel dimension to obtain key bidirectional temporal features of intraoperative infection. The intraoperative infection risk monitoring feature extraction is specifically carried out by performing temporal feature processing on key bidirectional temporal features of intraoperative infection based on a gated recurrent unit to obtain the hidden state of intraoperative infection temporal features. Then, through an attention mechanism, attention weights are calculated for the hidden state of intraoperative infection temporal features in each time window. Based on the attention weight values, all hidden states are weighted and aggregated to obtain the weighted fused intraoperative infection risk monitoring features. The infection monitoring results are output by performing a linear transformation on the intraoperative infection risk monitoring features through a fully connected layer, and then outputting the intraoperative operating room infection monitoring results through a Softmax activation function. Semi-supervised training of the model involves using intraoperative operating room infection management data from historical infection management optimization data as the training data for the model, designing a semi-supervised loss function based on the principle of semi-supervised learning, and performing semi-supervised training of the model to obtain the trained intraoperative operating room infection monitoring model. The semi-supervised loss function of the model is specifically a composite loss function that integrates supervised loss, consistency regularization loss, and pseudo-label loss. The supervised loss constrains the model prediction error of labeled data, the consistency regularization loss is used to ensure that the model's prediction results for unlabeled data remain consistent under perturbation changes, and the pseudo-label loss is used to guide the model to train on unlabeled data by generating pseudo-labels. Real-time monitoring of intraoperative operating room infection risk involves inputting intraoperative operating room infection management data from real-time infection management optimization data into a trained intraoperative operating room infection monitoring model to obtain real-time intraoperative operating room infection risk monitoring results.

[0009] Furthermore, the postoperative operating room infection terminal assessment module specifically includes the following steps: A postoperative operating room infection terminal assessment model was established. Specifically, a postoperative operating room infection terminal assessment model was established based on a multilayer perceptron neural network. Postoperative operating room infection management data from historical infection management optimization data was used as training data for the model to train the terminal assessment model and obtain the trained postoperative operating room infection risk assessment model. The final evaluation model hyperparameter adjustment is specifically based on the model hyperparameter optimization algorithm to obtain the postoperative operating room infection risk assessment model, adjust the hyperparameters of the trained postoperative operating room infection risk assessment model, and obtain the optimal postoperative operating room infection risk assessment model. The terminal real-time assessment of postoperative operating room infection involves inputting postoperative operating room infection management data from the real-time infection management optimization data into the optimal postoperative operating room infection risk assessment model to obtain the real-time terminal assessment result of postoperative operating room infection.

[0010] Furthermore, the intelligent management module for operating room infection specifically makes intelligent decisions on the infection status of the operating room based on real-time preoperative operating room infection risk assessment results, real-time intraoperative operating room infection risk monitoring results, and real-time postoperative operating room infection terminal assessment results, thereby realizing automated, intelligent, and closed-loop management of the operating room.

[0011] The beneficial effects achieved by the present invention using the above solution are as follows: (1) In view of the fact that traditional operating room infection management is mostly single-link control, lacking a complete process system of preoperative risk prediction, intraoperative dynamic control and postoperative final review, and has technical problems of insufficient accuracy in operating room infection risk assessment and monitoring, this solution innovatively combines the preoperative operating room infection risk assessment module, the intraoperative operating room infection dynamic monitoring module, and the postoperative operating room infection final assessment module to form a full-process, closed-loop infection risk management system. This ensures that the infection risk at each stage can be warned in real time, effectively controlled and rectified in a timely manner, thereby realizing dynamic monitoring and full-process control of infection risk, improving the early warning capability of infection risk, and enhancing the dynamism and real-time nature of operating room infection management. It can promptly detect and deal with potential risks at each stage of surgery, improve the comprehensiveness of operating room infection prevention and control, and realize intelligent and automated management of infection prevention and control.

[0012] (2) To address the technical problems of slow convergence speed, easy getting trapped in local optima, insufficient model hyperparameter optimization accuracy, poor model output accuracy and low evaluation efficiency in existing hyperparameter search algorithms applicable to evaluation models, this scheme innovatively introduces a piecewise convergence factor and an iterative back-learning optimization population strategy to improve the optimization algorithm. The piecewise nonlinear convergence factor can dynamically balance the global exploration and local development capabilities of the algorithm, accelerate the convergence process of the algorithm to the optimal hyperparameter combination, and the iterative back-learning population optimization strategy can continuously improve the population diversity throughout the algorithm's entire iteration cycle, significantly enhance the global optimization capability and iterative stability, improve the efficiency and reliability of hyperparameter search, thereby improving the model hyperparameter optimization accuracy and the accuracy of model output results, and finally realizing efficient decision-making for intelligent management of operating room infections.

[0013] (3) To address the technical problems of existing models for monitoring intraoperative operating room infections, which rely solely on the extraction of single temporal features, lack effective bidirectional temporal dependency capture, and are overly dependent on labeled data, while the amount of unlabeled data in the operating room is large and difficult to utilize, resulting in poor accuracy, real-time performance, and generalization ability of the infection monitoring model output, this solution innovatively proposes to design a bidirectional convolutional path and combine it with a gated recurrent unit. At the same time, it adopts a semi-supervised loss function that integrates supervised loss, consistency regularization loss, and pseudo-label loss for semi-supervised training of the model. This improves the ability to extract temporal features in the risk of operating room infection during the surgical process, breaks through the dependence of traditional models on fully labeled data, solves the problem of unlabeled data not being fully utilized in operating room infection monitoring, significantly improves the accuracy and real-time performance of the monitoring model, enhances the comprehensiveness and stability of infection risk monitoring, improves the robustness and generalization ability of the model, and finally realizes intelligent and automated operating room infection monitoring. Attached Figure Description

[0014] Figure 1 A schematic diagram of the modules of the operating room infection management system based on big data provided by the present invention; Figure 2 This is a module for preoperative operating room infection risk assessment. Figure 3 A flowchart illustrating the process of constructing the hyperparameter optimization algorithm for the preoperative operating room infection risk assessment module; Figure 4 A flowchart illustrating the process of constructing an intraoperative operating room infection monitoring model within the intraoperative operating room infection dynamic monitoring module; The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. Detailed Implementation

[0015] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0016] In the description of this invention, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the system or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0017] Example 1, see Figure 1 The operating room infection management system based on big data provided by the present invention includes a data acquisition module, a preoperative operating room infection risk assessment module, an intraoperative operating room infection dynamic monitoring module, a postoperative operating room infection terminal assessment module, and an operating room infection intelligent management module. The data acquisition module obtains optimized operating room infection management data through data collection and data preprocessing operations, and sends the data to the preoperative operating room infection risk assessment module, the intraoperative operating room infection dynamic monitoring module, and the postoperative operating room infection terminal assessment module. The preoperative operating room infection risk assessment module receives data sent by the data acquisition module and is used to conduct a comprehensive infection risk assessment of the operating room before surgery. Specifically, it establishes a preoperative operating room infection risk assessment model and introduces a piecewise convergence factor and an iterative back learning optimization population improvement algorithm to construct a model hyperparameter optimization algorithm. The optimal hyperparameter combination of the model is obtained by using this optimization algorithm to adjust the model hyperparameters and obtain the optimal preoperative operating room infection risk assessment model. Finally, the preoperative operating room infection management data is input into the model to obtain the real-time preoperative operating room infection risk assessment result, and the data is sent to the operating room infection intelligent management module. The intraoperative operating room infection dynamic monitoring module receives data sent by the data acquisition module and is used to dynamically monitor the infection risk in the operating room during surgery. Specifically, it extracts key bidirectional temporal features of intraoperative infection by designing forward and backward convolutional paths, and extracts intraoperative infection risk monitoring features by combining gated recurrent units and attention mechanisms. Finally, it outputs the infection monitoring results to construct an intraoperative operating room infection monitoring model. Then, it designs a semi-supervised loss function that integrates supervised loss, consistency regularization loss, and pseudo-label loss as the loss function for model training, and performs semi-supervised training of the model to obtain the trained intraoperative operating room infection monitoring model. Finally, it inputs intraoperative operating room infection management data into the monitoring model to obtain real-time intraoperative operating room infection risk monitoring results, and sends the data to the operating room infection intelligent management module. The postoperative operating room infection terminal assessment module receives data sent by the data acquisition module to assess the cleaning and disinfection effect of the operating room after surgery. Specifically, it establishes a postoperative operating room infection terminal assessment model and trains the model. At the same time, it uses a model hyperparameter optimization algorithm to obtain the optimal hyperparameter combination of the model and adjusts the model hyperparameters to obtain the optimal postoperative operating room infection risk assessment model. Finally, it inputs postoperative operating room infection management data into the model to obtain real-time postoperative operating room infection terminal assessment results and sends the data to the operating room infection intelligent management module. The intelligent management module for operating room infection receives data from the preoperative operating room infection risk assessment module, the intraoperative operating room infection dynamic monitoring module, and the postoperative operating room infection terminal assessment module. Specifically, it makes intelligent decisions on the infection status of the operating room based on the real-time results before, during, and after the operation.

[0018] By implementing the above operations, this solution addresses the technical issues of traditional operating room infection management, which often involves single-stage control and lacks a coherent system encompassing preoperative risk prediction, intraoperative dynamic control, and postoperative final review. This results in insufficient accuracy in operating room infection risk assessment and monitoring. This innovative solution combines a preoperative operating room infection risk assessment module, an intraoperative operating room infection dynamic monitoring module, and a postoperative operating room infection final assessment module to form a complete, closed-loop infection risk management system. This ensures that infection risks at each stage can be warned in real time, effectively controlled, and promptly rectified, thereby achieving dynamic monitoring and full-process control of infection risks. It improves the early warning capability for infection risks and enhances the dynamism and real-time nature of operating room infection management. It enables timely detection and handling of potential risks at each stage of surgery, improves the comprehensiveness of operating room infection control, and achieves intelligent and automated management of infection control.

[0019] Example 2, see Figure 1This embodiment is based on the above embodiment. Specifically, the data acquisition module obtains raw data of operating room infection management by performing data acquisition operations, and performs data preprocessing on the raw data of operating room infection management to obtain optimized data of operating room infection management. The raw data for operating room infection management includes historical infection management data and real-time infection management data; Both the historical infection management data and the real-time infection management data include preoperative operating room infection management data, intraoperative operating room infection management data, and postoperative operating room infection management data; The historical infection management data also includes historical operating room infection management results; The historical operating room infection management results include preoperative operating room infection risk assessment results, intraoperative operating room infection monitoring results, and postoperative operating room infection terminal assessment results. The preoperative operating room infection risk assessment results, intraoperative operating room infection monitoring results, and postoperative operating room infection terminal assessment results are all divided into four levels: low risk, moderate risk, high risk, and very high risk. The preoperative operating room infection management data includes cleanliness level, purification method, air supply method, preoperative time window pressure difference value, duration of preoperative time window pressure difference exceeding limit, preoperative time window temperature value, preoperative time window humidity value, temperature and humidity, duration of temperature and humidity exceeding limit, purification unit operating status, purification unit fault status, alarm status, number of alarms, cleaning and disinfection start time, cleaning and disinfection end time, cleaning and disinfection operation duration, cleaning method, disinfection method, disinfectant name, disinfectant type, disinfectant contact time, sterile item name in instrument pack, instrument pack release status, sterilization batch number, sterilization method, sterilization parameter qualification mark, number of pressure difference exceeding limit in the past N days, number of temperature and humidity exceeding limit in the past N days, number of purification alarms in the past N days, number of final assessment failures in the past N days, number of re-disinfections in the past N days, number of retests in the past N days, and average rectification time in the past N days. The intraoperative operating room infection management data includes differential pressure value sequence, temperature value, humidity value sequence, temperature and humidity change rate difference value, duration of temperature and humidity exceeding limits, purification unit operating status, alarm level, alarm trigger time, number of door openings, duration of door openings, frequency of door openings, personnel entry and exit count, number of people in the room, duration of personnel exceeding limits, interval between consecutive procedures, intraoperative critical events, critical event type, surgical type, patient disease type, and surgical cleanliness level; The postoperative operating room infection management data includes: terminal cleaning and disinfection start time, terminal cleaning and disinfection end time, terminal cleaning and disinfection operation duration, terminal cleaning method, terminal disinfection method, disinfectant name, disinfectant type, disinfectant concentration ratio, disinfectant contact time, completion status of key points, terminal cleaning and disinfection shifts, post-terminal pressure difference value, post-terminal temperature value, post-terminal humidity value, air sampling time, air sampling point, air testing items, air testing values, air testing conclusion label, surface sampling time, surface sampling point, surface testing items, surface testing values, surface testing conclusion label, ATP value, water sample sampling time, water sample point, water sample testing values, disinfectant concentration testing time, and disinfectant testing value. The data preprocessing is used to preprocess the collected raw data of operating room infection management, specifically by performing data cleaning, normalization and feature selection on the raw data to obtain optimized operating room infection management data. The data cleaning is used to improve the integrity and consistency of the original data, specifically by filling in missing values ​​and identifying and removing outliers. The normalization process is used to convert data of different dimensions and types into a computable feature representation. Specifically, it normalizes the numerical data in the original data using the min-max normalization method, mapping it to a unified numerical range, and encodes the label data using the one-hot encoding method, so that all types of data are converted into numerical features that can be used for calculation. The feature selection is used to filter key features that are strongly correlated with the risk of operating room infection from the normalized feature set and reduce the interference of redundant features on model training. Specifically, it is based on correlation analysis to filter feature fields in preoperative operating room infection management data, intraoperative operating room infection management data and postoperative operating room infection management data. First, the correlation coefficient between each feature field and the target label is calculated, and features with correlation below a preset threshold are removed, thereby obtaining the feature set for preoperative operating room infection risk assessment, the feature set for intraoperative operating room infection dynamic monitoring, and the feature set for postoperative operating room infection final assessment. Finally, the feature sets are combined to obtain optimized operating room infection management data.

[0020] Example 3, see Figure 1 , Figure 2 and Figure 3 This embodiment is based on the above embodiment, and the preoperative operating room infection risk assessment module specifically includes the following steps: A preoperative operating room infection risk assessment model was established. Specifically, the model was built based on a multilayer perceptron neural network. Preoperative operating room infection management data from historical infection management optimization data was used as training data for the model to obtain the trained preoperative operating room infection risk assessment model. The historical infection management optimization data refers to standardized feature data obtained by preprocessing historical infection management data from the original operating room infection management data. The model training uses a multi-class cross-entropy loss function, and iteratively updates the weight matrix and bias parameters of the evaluation model through backpropagation algorithm and gradient descent optimization method. The evaluation model parameters are continuously optimized through multiple rounds of iteration. When the preset maximum number of training times is reached or the multi-class cross-entropy loss function value converges to a set threshold, the iterative training stops. Constructing a model hyperparameter optimization algorithm, specifically, obtaining the optimal combination of hyperparameters for the model through an improved optimization algorithm; including the following steps: Initialize the population search individuals to encode the core hyperparameters of the target model into searchable position vectors and generate an initial population covering a reasonable search space. Specifically, encode the model's hyperparameters into search individual position vectors and generate M search individual position vectors through random initialization. Each search individual position vector uniquely corresponds to a set of candidate individual model hyperparameter combinations, forming the initial search individual population. The hyperparameters of the model include the number of hidden layers, the number of neurons per layer, the type of activation function in the hidden layers, and the Dropout rate; The fitness calculation of the search individual specifically involves calculating the fitness value of the search individual in the population; the performance of the model built based on the hyperparameter combination corresponding to the search individual's position is used as the fitness value of the search individual. The piecewise convergence factor is calculated to divide the iteration stages by the split point. Combined with the coordinated control of the nonlinear and linear segments, it achieves a dynamic balance between strong exploration in the early stage and refined development in the later stage. Specifically, the split point between the nonlinear and linear segments is first calculated based on the number of iterations. Then, using the split point as the boundary, the piecewise convergence factor for the nonlinear exploration segment and the linear development segment are calculated separately. The formula used is as follows: ; ; ; ; In the formula, t represents the current iteration number. Indicates the maximum number of iterations. This represents the convergence control coefficient, with a range of values. , This represents the coefficient of the first-order term at the segmentation point. This represents the coefficient of the constant term at the segmentation point. This represents the split point in the t-th iteration. Denotes the piecewise convergence factor for the t-th iteration; The search individual position update is used to collaboratively determine the position update logic based on random numbers and search control parameters, and to execute the update logic according to different scenarios. This enables precise switching between global exploration and local development of the algorithm. Specifically, for each search individual, the search individual position is updated according to random numbers and search control parameters, based on different scenarios. The formula used is as follows: ; ; ; ; In the formula, Indicates search control parameters. This represents the distance between the currently searching individual and a randomly selected searching individual. This represents the distance between the current searched individual and the optimal searched individual. Indicates a random search for individuals. This indicates that the population is searching for the globally optimal position of an individual in the t-th iteration. and Indicates being between Random numbers uniformly distributed within a range This represents the position of the i-th individual in the d-th dimension within the t-th generation of the population. This represents the candidate position of the i-th individual in the d-th dimension within the (t+1)-th generation of the population. Indicates being between Random numbers uniformly distributed within a range This represents the parameter that controls the shape of the spiral search, and its value range is... , This represents the random spiral parameter, with a range of values. ; Iterative back-learning optimizes the population, overcoming the limitation of traditional back-learning being used only for initialization. At the end of each iteration, a back-learning operation is performed on the updated population. This improves population diversity by generating back-learning solutions and selectively choosing the best, thus avoiding the algorithm getting trapped in local optima. Specifically, for the updated search individuals, back-learning individuals are generated based on the search space boundary, resulting in a back-learning population. The original population and the back-learning population are then merged to obtain a mixed population of size 2M. Finally, the fitness values ​​of all search individuals in the mixed population are calculated, and after sorting by fitness, the M search individuals with the best fitness are selected as the population population. The initial population is iterated, and the individual with the best fitness value is selected for the search. The global optimum position of the search individual is then updated. The formula used is as follows: ; In the formula, This represents the reverse position of the i-th individual in the d-th dimension within the (t+1)-th generation of the population. This represents the lower bound of the searched individual at the d-th dimension. This represents the upper bound of the searched individual's position in the d-th dimension; The search iteration terminates when the fitness value corresponding to the global optimal position of the search individual is higher than the fitness threshold or when the number of iterations reaches the maximum number of iterations. The global optimal position of the search individual specifically refers to the optimal combination of hyperparameters of the model. Model hyperparameter tuning specifically involves obtaining the optimal hyperparameter combination of the preoperative operating room infection risk assessment model based on the model hyperparameter optimization algorithm, adjusting the hyperparameters of the trained preoperative operating room infection risk assessment model, and obtaining the optimal preoperative operating room infection risk assessment model. Real-time preoperative operating room infection risk assessment involves inputting preoperative operating room infection management data from real-time infection management optimization data into the optimal preoperative operating room infection risk assessment model to obtain real-time preoperative operating room infection risk assessment results. The real-time infection management optimization data is standardized feature data obtained by preprocessing the real-time infection management data in the original operating room infection management data.

[0021] By performing the above operations, this solution addresses the technical problems of existing hyperparameter search algorithms suitable for evaluating models, such as slow convergence speed, susceptibility to local optima, resulting in insufficient hyperparameter optimization accuracy, poor model output accuracy, and low evaluation efficiency. It innovatively introduces a piecewise convergence factor and an iterative back-learning optimization population strategy to improve the optimization algorithm. The piecewise nonlinear convergence factor dynamically balances the algorithm's global exploration and local exploitation capabilities, accelerating the convergence process towards the optimal hyperparameter combination. The iterative back-learning population optimization strategy continuously improves population diversity throughout the algorithm's entire iteration cycle, significantly enhancing global optimization ability and iterative stability, improving the efficiency and reliability of hyperparameter search, thereby increasing the model's hyperparameter optimization accuracy and the accuracy of model output results. Ultimately, this achieves efficient decision-making for intelligent management of operating room infections.

[0022] Example 4, see Figure 1 and Figure 4 This embodiment is based on the above embodiment, and the intraoperative operating room infection dynamic monitoring module specifically includes the following steps: The construction of an intraoperative operating room infection monitoring model includes the following steps: The extraction of key bidirectional temporal features for intraoperative infection is used to comprehensively mine bidirectional temporal correlation features related to intraoperative infection. This involves capturing both the positive temporal logic from preceding operative actions to subsequent infection risk and the negative temporal correlation from abnormal infection risk to tracing back to critical oversights in the early stages. Ultimately, core bidirectional temporal features strongly correlated with intraoperative infection are extracted. Specifically, a bidirectional convolutional neural network is designed to extract both positive and negative infection risk temporal features, and finally, bidirectional feature fusion is performed to obtain key bidirectional temporal features of intraoperative infection. The process includes the following steps: The design of the forward convolution path involves taking the dynamic monitoring feature set of intraoperative operating room infection as input, and performing sliding convolution calculations along the forward temporal direction of the operation time using a 3×3 convolution kernel to generate positive local feature values. After aggregation and dimensionality reduction by ReLU activation function and max pooling operation, positive infection risk temporal features that can reflect the transmission law of the preceding operation behavior to the subsequent infection risk are extracted. The design of the reverse convolution path involves flipping the intraoperative operating room infection dynamic monitoring feature set along the time axis, performing sliding convolution calculations using a 3×3 kernel along the flipped reverse temporal direction to generate reverse local feature values, aggregating and reducing the dimensionality using the ReLU activation function and max pooling operation to obtain reverse pooling features, flipping the reverse pooling features again along the time axis to restore the temporal order consistent with the original operation time, and extracting reverse infection risk temporal features that can support the retrospective analysis of key oversights in the early stages of infection risk anomalies. Bidirectional feature fusion is used to integrate the temporal features of positive and negative infection risks, mine the bidirectional correlation value between features, eliminate the information limitations of single-path features, and finally form core bidirectional temporal features that are dimensionally unified, informationally complete and strongly correlated with intraoperative infection. Specifically, the temporal features of positive and negative infection risks are dimensionally aligned and spliced ​​and fused in the feature channel dimension to obtain key bidirectional temporal features of intraoperative infection. Intraoperative infection risk monitoring feature extraction is used to improve the representation effectiveness of infection risk-related features by adaptively weighting and strengthening key temporal features strongly correlated with infection during surgery, weakening the interference of irrelevant temporal features. Specifically, it performs temporal feature processing on key bidirectional temporal features of intraoperative infection based on gated recurrent units to obtain the hidden state of intraoperative infection temporal features. Then, through an attention mechanism, it calculates the attention weight of the hidden state of intraoperative infection temporal features in each time window. Based on the attention weight value, it performs weighted aggregation processing on all hidden states to obtain the weighted fused intraoperative infection risk monitoring features. The infection monitoring results are output to quantify the probability of intraoperative infection risk based on the weighted fusion of the core temporal features of intraoperative infection risk. Specifically, the intraoperative infection risk monitoring features are linearly transformed through a fully connected layer, and the intraoperative operating room infection monitoring results are output through a Softmax activation function. Semi-supervised training of the model is used to fully utilize the labeled data of intraoperative operating room infection management and massive unlabeled historical data for model training. This breaks through the dependence of pure supervised training on a large amount of infection labeled data, improves the model's learning ability and generalization performance of intraoperative infection risk features, and makes the trained model more suitable for the actual scenario where clinical intraoperative infection labeled data is scarce. This ensures the predictive accuracy and robustness of the model in real-time intraoperative monitoring. Specifically, the intraoperative operating room infection management data in the historical infection management optimization data is used as the training data of the model, and a semi-supervised loss function is designed based on the semi-supervised learning principle to conduct semi-supervised training of the model, resulting in the trained intraoperative operating room infection monitoring model. The semi-supervised loss function of the model is specifically a composite loss function that integrates supervised loss, consistency regularization loss, and pseudo-label loss. The supervised loss constrains the model's prediction error on labeled data. The consistency regularization loss ensures the prediction stability of unlabeled data under input perturbations. By slightly perturbing the unlabeled data, the model's output should remain consistent, thus forcing the model to learn more robust feature representations and enhancing its adaptability and generalization ability to unlabeled data. The pseudo-label loss guides the model's training on unlabeled data by generating pseudo-labels. By generating high-confidence predictions on unlabeled data and incorporating these predictions as pseudo-labels into the loss function for optimization, the model can better learn potential patterns and structures from unlabeled data while avoiding noise interference from pseudo-labels, achieving collaborative training of labeled data, ordinary unlabeled data, and high-confidence unlabeled data. The formula used is as follows: ; ; ; ; In the formula, This represents the value of the composite loss function. This indicates the value of the supervised loss item. This represents the value of the consistency regularization loss term. This represents the value of the pseudo-label loss term. The balance coefficient represents the consistency regularization loss term, and its value ranges from 1 to 2. , This represents the balance coefficient of the pseudo-label loss term, with a value range of [value missing]. , This indicates the sample size of the training data for intraoperative infection markers. This represents the total number of categories of intraoperative infection risk types. This indicates that the i-th labeled sample corresponds to the true label of the c-th class. This represents the predicted probability that the model believes the i-th labeled sample belongs to the c-th class. This represents the sample size of routine, unlabeled training data during surgery, i.e., the number of routine surgical data entries without a clear infection risk level label. This represents the probability vector of the model's predicted infection risk level for the j-th ordinary unlabeled sample. This represents the model's predicted probability vector for the infection risk level of the j-th ordinary unlabeled sample after adding a slight Gaussian noise perturbation. This represents the square operation of the L2 norm. This represents the number of unlabeled training data samples for which the model's prediction confidence is higher than a set threshold. This indicates that the pseudo-label of the k-th high-confidence unlabeled sample corresponds to the c-th class. This represents the model's predicted probability that the k-th high-confidence unlabeled sample belongs to class c; The semi-supervised training of the model adopts a semi-supervised loss function and uses backpropagation algorithm and gradient descent optimization method to iteratively update the weight matrix and bias parameters of the monitoring model. The model parameters are continuously optimized and evaluated through multiple rounds of iteration. When the preset maximum number of training times is reached or the composite loss function value converges to a set threshold, the iterative training stops. Real-time monitoring of intraoperative operating room infection risk involves inputting intraoperative operating room infection management data from real-time infection management optimization data into a trained intraoperative operating room infection monitoring model to obtain real-time intraoperative operating room infection risk monitoring results.

[0023] By performing the above operations, this solution addresses the technical problems of existing models for intraoperative operating room infection monitoring, which rely solely on the extraction of single temporal features, lack effective bidirectional temporal dependency capture, and over-rely on labeled data, while the large amount of unlabeled data in the operating room is difficult to utilize, resulting in poor accuracy, real-time performance, and generalization ability of the infection monitoring model output. This solution innovatively proposes a bidirectional convolutional path combined with a gated recurrent unit, and employs a semi-supervised loss function that integrates supervised loss, consistency regularization loss, and pseudo-label loss for semi-supervised model training. This improves the ability to extract temporal features from operating room infection risk during surgery, overcomes the dependence of traditional models on fully labeled data, solves the problem of underutilization of unlabeled data in operating room infection monitoring, significantly improves the accuracy and real-time performance of the monitoring model, enhances the comprehensiveness and stability of infection risk monitoring, and improves the robustness and generalization ability of the model, ultimately achieving intelligent and automated operating room infection monitoring.

[0024] Example 5, see Figure 1 This embodiment is based on the above embodiment, and the postoperative operating room infection terminal assessment module specifically includes the following steps: A postoperative operating room infection terminal assessment model was established. Specifically, a postoperative operating room infection terminal assessment model was established based on a multilayer perceptron neural network. Postoperative operating room infection management data from historical infection management optimization data was used as training data for the model to train the terminal assessment model and obtain the trained postoperative operating room infection risk assessment model. The final evaluation model training adopts the multi-class cross-entropy loss function, and iteratively updates the weight matrix and bias parameters of the evaluation model through backpropagation algorithm and gradient descent optimization method. The evaluation model parameters are continuously optimized through multiple rounds of iteration. When the preset maximum number of training times is reached or the multi-class cross-entropy loss function value converges to a set threshold, the iterative training stops. The final evaluation model hyperparameter adjustment is specifically based on the model hyperparameter optimization algorithm to obtain the postoperative operating room infection risk assessment model, adjust the hyperparameters of the trained postoperative operating room infection risk assessment model, and obtain the optimal postoperative operating room infection risk assessment model. The terminal real-time assessment of postoperative operating room infection involves inputting postoperative operating room infection management data from the real-time infection management optimization data into the optimal postoperative operating room infection risk assessment model to obtain the real-time terminal assessment result of postoperative operating room infection.

[0025] Example 6, see Figure 1 This embodiment is based on the above embodiment. Specifically, the intelligent management module for operating room infection is used to make intelligent decisions on the infection status of the operating room based on the real-time preoperative operating room infection risk assessment results, the real-time intraoperative operating room infection risk monitoring results, and the real-time postoperative operating room infection terminal assessment results, so as to realize the automated, intelligent and closed-loop management of the operating room. The intelligent decision-making specifically involves automatically generating targeted management measures based on the infection risk level of real-time results to ensure that the operating room meets infection control requirements.

[0026] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0027] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention.

[0028] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. A big data-based operating room infection management system, characterized by: It includes a data acquisition module, a preoperative operating room infection risk assessment module, an intraoperative operating room infection dynamic monitoring module, a postoperative operating room infection terminal assessment module, and an operating room infection intelligent management module; The data acquisition module is used to obtain optimized data for operating room infection management; The preoperative operating room infection risk assessment module is used to establish a preoperative operating room infection risk assessment model. It adopts an improved optimization algorithm that introduces a piecewise convergence factor and an iterative back learning optimization population strategy to obtain the optimal hyperparameter combination of the model and adjust the model hyperparameters. Based on the model after hyperparameter adjustment, the real-time preoperative operating room infection risk assessment result is obtained. The intraoperative operating room infection dynamic monitoring module is used to extract key bidirectional temporal features of intraoperative infection through bidirectional convolutional paths, and to construct an intraoperative operating room infection monitoring model by combining gated recurrent units and attention mechanisms. The model is semi-supervised by using a model semi-supervised loss function that integrates supervised loss, consistency regularization loss, and pseudo-label loss. Based on the trained model, the real-time intraoperative operating room infection risk monitoring results are obtained. The postoperative operating room infection terminal assessment module is used to obtain real-time postoperative operating room infection terminal assessment results; The intelligent management module for operating room infection is used to intelligently manage the infection status of the operating room based on real-time results before, during, and after surgery.

2. The operating room infection management system based on big data according to claim 1, characterized in that: The preoperative operating room infection risk assessment module specifically includes the following steps: A preoperative operating room infection risk assessment model was established, specifically based on a multilayer perceptron neural network. The preoperative operating room infection management data from historical infection management optimization data was used as the training data for model training. Construct a model hyperparameter optimization algorithm; Model hyperparameter tuning specifically involves obtaining the optimal hyperparameter combination of the preoperative operating room infection risk assessment model based on the model hyperparameter optimization algorithm, adjusting the hyperparameters of the trained preoperative operating room infection risk assessment model, and obtaining the optimal preoperative operating room infection risk assessment model. Real-time preoperative operating room infection risk assessment involves inputting preoperative operating room infection management data from real-time infection management optimization data into the optimal preoperative operating room infection risk assessment model to obtain real-time preoperative operating room infection risk assessment results.

3. The operating room infection management system based on big data according to claim 2, characterized in that: The hyperparameter optimization algorithm for the constructed model specifically includes the following steps: Initialize the population of search individuals by encoding the model's hyperparameters into search individual position vectors and generating M search individual position vectors through random initialization to form the initial search individual population; The fitness calculation of individual search involves calculating the fitness value of individual search in the population, using the model performance built based on the hyperparameter combination corresponding to the position of the individual search as the fitness value of the individual search. To calculate the piecewise convergence factor, firstly, the split point between the nonlinear and linear segments is calculated based on the number of iterations, and then the piecewise convergence factor is calculated using the split point as the boundary. The search involves updating the location of each individual being searched. Iterative back learning optimizes the population; The search iteration terminates when the fitness value corresponding to the global optimal position of the search individual is higher than the fitness threshold or when the number of iterations reaches the maximum number of iterations. The global optimal position of the search individual specifically refers to the optimal combination of hyperparameters of the model.

4. The operating room infection management system based on big data according to claim 3, characterized in that: The iterative back-learning optimization of the population specifically involves generating a back-search population based on the search space boundary for the updated search individuals. The original population and the back-search population are then merged to obtain a mixed population of size 2M. Finally, the fitness values ​​of all search individuals in the mixed population are calculated, and after sorting them by fitness, the M search individuals with the best fitness are selected as... The initial population is iterated, and the individual with the best fitness value is selected for the search, and the global optimal position of the search individual is updated.

5. The operating room infection management system based on big data according to claim 1, characterized in that: The intraoperative operating room infection dynamic monitoring module specifically includes the following steps: The construction of an intraoperative operating room infection monitoring model includes the following steps: The key bidirectional temporal features of intraoperative infection were extracted by designing a bidirectional convolutional neural network to extract the temporal features of positive infection risk and negative infection risk, and finally the bidirectional features were fused to obtain the key bidirectional temporal features of intraoperative infection. Intraoperative infection risk monitoring feature extraction is specifically achieved by processing key bidirectional temporal features of intraoperative infection based on a gated recurrent unit and combining them with an attention mechanism to obtain intraoperative infection risk monitoring features. The infection monitoring results are output specifically through a fully connected layer and a Softmax activation function to output the intraoperative operating room infection monitoring results. Semi-supervised training of the model involves using intraoperative operating room infection management data from historical infection management optimization data as the training data for the model, and designing a semi-supervised loss function based on the principle of semi-supervised learning to perform semi-supervised training of the model, thereby obtaining the trained intraoperative operating room infection monitoring model; the semi-supervised loss function is specifically a composite loss function that integrates supervised loss, consistency regularization loss, and pseudo-label loss. Real-time monitoring of intraoperative operating room infection risk involves inputting intraoperative operating room infection management data from real-time infection management optimization data into a trained intraoperative operating room infection monitoring model to obtain real-time intraoperative operating room infection risk monitoring results.

6. The operating room infection management system based on big data according to claim 5, characterized in that: The design of the forward convolution path specifically involves taking the dynamic monitoring feature set of intraoperative operating room infection as input, performing sliding convolution calculation along the forward temporal direction of the operation time, and obtaining the forward infection risk temporal features through activation function and max pooling operation; The design of the reverse convolution path specifically involves performing sliding convolution calculations on the intraoperative operating room infection dynamic monitoring feature set along the reverse temporal direction of the flip, and obtaining reverse pooling features through activation functions and max pooling operations. The reverse pooling features are then flipped again along the time axis to obtain reverse infection risk temporal features.

7. The operating room infection management system based on big data according to claim 1, characterized in that: The postoperative operating room infection terminal assessment module specifically establishes a postoperative operating room infection terminal assessment model based on a multilayer perceptron neural network, trains the terminal assessment model, then obtains a postoperative operating room infection risk assessment model based on a model hyperparameter optimization algorithm, adjusts the hyperparameters of the trained postoperative operating room infection risk assessment model, performs terminal assessment model hyperparameter adjustment, and finally inputs postoperative operating room infection management data from real-time infection management optimization data into the model after hyperparameter adjustment to obtain real-time postoperative operating room infection terminal assessment results.

8. The operating room infection management system based on big data according to claim 1, characterized in that: The intelligent management module for operating room infection specifically makes intelligent decisions regarding the infection status of the operating room based on real-time preoperative operating room infection risk assessment results, real-time intraoperative operating room infection risk monitoring results, and real-time postoperative operating room infection terminal assessment results.

9. The operating room infection management system based on big data according to claim 1, characterized in that: The data acquisition module specifically obtains raw operating room infection management data by performing data acquisition operations, and preprocesses the raw operating room infection management data to obtain optimized operating room infection management data; the raw operating room infection management data includes historical infection management data and real-time infection management data.

Images